Redundant shape or pattern coding alongside colour

What it is

Categorical encoding expresses the same category variable through two channels at once, colour plus shape/pattern: each line in a line chart carries both a colour and a marker (circle, square, triangle), scatter points are both tinted and shaped, map regions both filled and hatched. Redundancy is not waste — it double-insures category identity: colour-deficient readers take the shape channel, black-and-white printing takes the pattern channel, small marks where colour fades still carry shape, and near-neighbour colour pairs (when the palette crowds its ceiling) separate by shape. Colour is the fast channel, shape the reliable one; redundancy lets them cover each other.

Why it happens

Colour and shape failure modes do not overlap: colour degrades under colour-vision deficiency, greyscale media, small areas, and glare; shape survives all of these, weak only in smaller capacity (about 4–6 quickly distinguishable markers) and slightly slower retrieval (colours pop in parallel; shape search leans serial). Dual-channel stacking raises each channel's shortest plank: near-neighbour colours separate by shape, confusable shapes (circle versus ellipse) separate by colour; legend learning accelerates too — the two-attribute binding "red circle" is sturdier than either alone. The only cost is a modest rise in visual density, and that cost concentrates exactly in dense charts, which is precisely where colour is least reliable — net positive by arithmetic.

Where it stops holding

Redundant capacity is the intersection of the two channels: about 6 shapes, about 10 colours, a practical redundant ceiling around 6 categories, beyond which return to aggregation (previous leaf). Markers on line charts mean something only at sparse points (dense points stack markers into mush — fall back to pure colour plus facets); pattern fills (hatching, stippling) work for large areas, not small marks. Emphasis redundancy (red plus bold) differs from categorical redundancy — that is importance on two channels, not category identity; keep them apart.

Applying it

• Default templates to "one colour, one shape" for line charts (a 6-shape cycle over 6 colours), same rule for scatterplots; past 6 categories, aggregate before abandoning shape.
• Choose marker sets with large separations (circle, square, triangle, cross, diamond, ×), avoiding orientation-dependent shapes (ellipse versus circle).
• Verification: run category identification after greyscale printing or colour-blindness simulation (desaturation filter); the redundant version should hold while the colour-only version's drop measures the redundancy's value.

Related

• Same group: U4.01.1 The number of distinguishable categories has a ceiling · U4.01.2 Group or aggregate when categories multiply
• Nearby: U1.10.1 Redundant covariation across channels can speed reading · U4.04.1 Red-green pairs fail for a sizable population
• Search terms: redundant encoding · shape markers · dual coding · colour blindness safe